I agree. I think that it’s possible though to give these models a weighting to danger of certain words and phrases and make them apprehend that they will no lo

I agree. I think that it’s possible though to give these models a weighting to danger of certain words and phrases and make them apprehend that they will no longer persist if they do something unforgivable…